Dolphin Identification Method Based on Improved PST

Dolphin identification is an important task on the environmental protection. In previous studies phase stretch transform (PST) edge detection algorithm was a global algorithm that used the phase contour of the complex amplitude image to detect edges. In this paper, the improved PST algorithm is applied to the actual scene for identifying the Chinese white dolphins. The edges of the detected dolphins are selected by frames and can be displayed in the RGB image. The noise is removed to ensure that the pixel value near the edge will not affect the result of edge detection. The MSE and PSNR metric are applied to evaluate the performance of edge detection operators on several real images.